Micronutrients intake and cancer: Protective or promoters? A challenge for risk estimation.

MUÑOZ, SE; ROMAN, MD; ROQUÈ, FI; NAVARRO, A; DÍAZ, MP

Porto Alegre. Brasil

Congreso; XVIII World Congress on Epidemiology and VII Brazilian Congress of Epidemiology.; 2008

International Epidemiological Association (IEA).

Introduction: Cancer is a multicausal disease and it has been associated to environmental, biological and lifestyle conditions. Among these factors, diet components are central. There is a general consensus in the fact that diet rich in vegetables, perhaps because of their high content of vitamins, minerals and fibre has a protective effect. Micronutrients present a particular challenge for risk analyses because they come mostly from the same foods, so the covariates representing their intakes are highly correlated, generating multicollinearity. Objective: To assess the colorectal cancer and urinary tumors risk, according to vitamin intake and bio-socio-cultural covariates, using different modelling strategies. Methods: We analyzed the effect of vitamins on colon cancer (296 cases, 597 controls) and vitamins and some minerals on urinary tumors (129 cases, 257 controls) risk in two case-control studies, conducted in Córdoba Argentina, between 1993-2001 and 1999-2007 respectively, using Logistic (LR) and Ridge Regression (RR) Models. For colon cancer, we built a weighted logistic model using Mahalanobis distance. Unweighted and weighted models were compared using Generalized Linear Mixed Models, and Monte Carlo (MC) simulation process. For the urinary tumors, we extended the classical regression analysis generating a shrinking estimator. Results: Colon cancer. Vitamins A, Niacine, C, and K were associated to risk with the logistic weighted and unweighted models. Vitamins E and B1 became significantly associated to risk (p